Principles for an Implementation of a Complete CT Reconstruction Tool Chain for Arbitrary Sized Data Sets and Its GPU...

Area of Science:

• Medical Imaging
• Computer Science
• Image Processing

Background:

• Computed tomography (CT) reconstruction is computationally intensive.
• Efficient reconstruction algorithms are crucial for clinical and research applications.
• Existing frameworks may lack flexibility or speed for specific tasks.

Purpose of the Study:

• To describe the implementation of an efficient and fast in-house computed tomography (CT) reconstruction framework.
• To detail the core filtered backprojection algorithm and its optimization on Graphics Processing Units (GPUs).
• To present methods for artifact reduction and precise geometric correction in CT imaging.

Main Methods:

• Implementation of a cone-beam CT reconstruction toolchain.
• Optimization of filtered backprojection using GPU hardware acceleration.
• Development of algorithms for artifact reduction (ring artifacts, beam hardening) and geometric correction (center of rotation, tilted axis).
• Application of data splitting and GPU kernel optimization strategies for large datasets.

Main Results:

• An efficient and fast in-house CT reconstruction framework was successfully implemented.
• The framework supports the reconstruction of CT images of arbitrary data size.
• GPU acceleration significantly speeds up the filtered backprojection process.
• Methods for effective artifact reduction and precise geometric correction were integrated.

Conclusions:

• The developed framework offers an efficient and versatile solution for CT image reconstruction.
• GPU optimization is key to achieving high-speed performance in CT reconstruction.
• The integrated artifact reduction and correction tools enhance image quality and reliability.